[mnkiefer]

Refactors server/internal/adxtrace/capture.go so per-run evidence classification lives in a testable captureCandidate function, and adds one debug summary log (cao:adxtrace) per capture pass. Behavior is unchanged.

Action: Reviewer should confirm the extraction is behavior-preserving and that go -C server vet ./... && go -C server test ./... passes.

Change

• CaptureEvidence loop body moved into captureCandidate, which returns evidence, an exclusion reason, or an I/O error. Reason strings and placeholder evidence are identical to before.
• One cacheLog.Printf after a completed pass logs only counts: candidates, projected, excluded. Disabled by default. No identifiers, paths, errors, or content are logged, and it is outside the per-candidate loop.

Tests

New capture_test.go uses real temp directories and real projection (no mocks) to cover projected, identity-gap, directory-gap, unavailable, invalid-source and size-exclusion outcomes, plus out-of-scope repository skipping.

Validation

gofmt clean; go -C server vet ./... and go -C server test ./... pass.

Control Plane

• Correlation ID: 41829643232-3314
• Central repository: githubnext/gh-cao
• Control plane run: https://github.com/githubnext/gh-cao/actions/runs/41829643232

Generated by SelfCare / Server Go Logging · copilot · auto · ~34 AIC · ⌖ ~1 AIC · ⊞ 14.9K · ◷

• expires on Nov 11, 2026, 12:45 PM UTC

[mnkiefer]

@copilot Fix failing checks

[Copilot]

@copilot Fix failing checks

Fixed the nilerr finding in b8a1ec3 by separating projection classification from I/O error handling. golangci-lint fmt --diff, golangci-lint run ./..., go vet ./..., and go test ./... pass locally; CodeQL found no alerts.